Artificial Intelligence Drives the Archaeological Discoveries of Stone Ageby Astha Oriel October 27, 2020
Machine learning model and Neural Networks helps in extracting archaic information about human civilization.
Archaeology is the gateway to our past. It describes events which shaped the world how it is today and the transition that led humans from animal-hunter to a knowledgeable-mosaic. In archaeology, Stone Age holds the key relevance. It establishes the patterns of human behavior and helps in identifying the transitions that hurled humans to the path of development. It was also the era which disembarked human’s acquaintance with tools to sustain their living. Stone Age displayed the creative and clever aspect of human civilization.
Over the years, researchers and scientists are keen to discover the incidents of the Stone Age, which led to the expansion of human civilization. They are inquisitive to learn and discover about how civilizations travel from one place to other and resettled themselves. In a traditional research setting, the information retrieved from many sources is not sufficient to arrive at a substantive conclusion. Henceforth, researchers and scientists are leveraging machine learning model to determine their archaeological findings.
Clarifying the Archaeological Origin
A team of Mexican archaeologists and the University of Marburg are leveraging machine learning model to identify whether the source material required for making Obsidian artifacts, discovered in Xalasco came from local sources or were obtained from other remote areas. Xalasco is a pre-Colombian cite in western Mexico.
In a paper titled“Projection-Based Classification of Chemical Groups for Provenance Analysis of Archaeological Materials”, the researcher states that with the combination of unsupervised and semi-supervised machine learning and chemometric application on the samples of Mesoamerican geological sources and obsidian artifacts collected from the archaeological site of Xalasco in Mexico a preference of Xalasco inhabitants to local obsidian deposits have become evident.
The paper also pointed out that with the combination of XRF spectroscopy, an adequate tool for obsidian provenance studies and which can successfully discriminate between several groups of archaeological artifacts employing quantitative analysis, and machine learning algorithms aided the researchers to implement a procedure that automatically determines the number of groups in obsidian samples according to their chemical characteristics and, consequently, defines their origin.
Moreover, with the machine learning method, which became a flexible and robust approach for cluster analysis consisting of three separate modules that can be optionally combined into the Databionic Swarm (DBS), the non-linear projection displayed the structure of the high-dimensional data into a low-dimensional space preserving the cluster structure of the data. It must be noted that the Databionic Swarm is part of the Swarm-based projection method, which assist intelligent agents to interact with each other and with the environment by displaying intelligent behavior. Swarm intelligence is considered as the most fertile ground for establishing new methodologies of classification.
The paper remarks DBS to be an efficient automatic method which allows determining the number of clusters that can be deduced from the topographic map of the Pswarm projection. Pswarm Projection also known as P-swarm is the swarm projection of high-dimensional data onto a two-dimensional plane by using intelligent agents operating on a toroidal and polar grid.
The use of DBS and the generalized U-matrix provides apparent advantages to classical clustering algorithms. The U-matrix, or Unified Distance matrix, is a method used for the self-organizing map (SOM) in which the Euclidian distance between the codebook vectors of neighbouring neurons define the height of each position of a codebook.
The use of the generalized U-matrix allows visually validating the cluster structures directly in the data by the visualization of the topographic map, without the help of additional analysis or plots. If a high dimensional dataset contains distance and/or density-based clusters, the resulting topographic map landscape possesses a clear valley – mountain ridge – valley structure.
In the topographic map, the borders of the visualization are cyclically connected with a periodicity and assists in the correct estimation of the number of clusters, by assessing them. The clustering is valid if mountain ranges do not partition the clusters, indicated by coloured points of the same colour and regions of points. In the 3D visualization of the 3D topographic map, the observations with similar characteristics are represented as valleys, whereas the differences are represented as mountain ranges and the outliers are seen as volcanos.
The paper states that the above-mentioned approach helps discover the origin of the artifacts.
For Differentiating between Tools sets between stone and Middle Age
Another utilization of AI-technique to assist in archaeological findings is in differentiating between different tools in middle to the later Stone Age. A team of researchers from the University of Liverpool and the Max Planck Institute for the Human History Jena, have leveraged machine learning neural networks for this.
With the help of this method, the team remains hopeful of identifying the various cultural difference of east Africa.
Archaeology consists of numerous unexplored entities, areas and information that are yet to be reached out by researchers and scientists. A traditional method has a parochial scope for extracting such information. Thus, by integrating machine learning models and AI-subsets, the process of excavating and discovering new information about human civilization would become easier.